Differential expression pattern-based prioritization of candidate genes through integrating disease-specific expression data

Expression data can reveal subtle transcriptional changes that mediate the clinical phenotype of the disease resulting from interaction between genetic and environmental factors, which offers us a new perspective to prioritize candidate genes. Here, we proposed a novel differential expression pattern (DEP)-based approach integrating numerous disease-specific expression data sets for prioritizing candidate genes. Using breast cancer as a case study, we validated the efficiency of our approach through integrating 12 breast cancer-related expression data sets based on the leave-one-out cross-validation. Particularly, prioritization based on subtype-specific expression data sets could generate significantly higher performance. The performance could be continually improved with the increasing expression data sets regardless of platform heterogeneity. We further validated the robustness of this approach by application to prostate cancer. Additionally, our approach showed higher performance in comparison with other expression-based approaches and better capability of identification of less well-studied disease genes in comparison with other integration-based approaches.